NS: investigated SBO (Systems Biology Ontology)-- using content markup to mark variables and relationships between variables in different models; MathML deeply embedded in there; it's clearly a part of it, but limited to mathematical aspects and not to biological equations
GJR: do they make proper use of MathML?
NS: yes, from what i saw, they are using properly - using content side to convey semantics - done very well; just one part of whole model
NS: quite a large collection - 250 models; active group - European funding
GJR: any SVG or SVG use?
NS: wasn't looking for it, but didn't run across it in my wandering through the site
NS: what should we get out of exercise
GJR: first: are XML syntax rules being observed and proper markup, where extant (such MathML) is employed; second thing: how are they exposing the content -- how is it being rendered? is it renderable by AT? is there a way to harness the default rendering, or must alternates be explored?
NS: basically the SBO-related sites you pointed us to have giant database with XML behind it which describes molecules and proteins - relationships between them can be graphically display molecule, lots of roles and properties; graphical, tabular and hyperlinks; graph, then table with ton of links in table for properties of molecule;
NS: not going to see in web page
NS: don't see how can deliver as raw HTML - too complicated
GJR: any special plugins needed for display?
NS: put up web page with forms;
GJR: back-end relational database?
GJR: so scenario is serve most appropriate from backplane, rather than interpret?
NS: may have been done because browsers not able to handle
NS: expert handlers may come up against too much info;
MC: RixML covering domains in general way; MathML is specific
MC: apply expert handler to sub-content - don't know if EH for explicit language -- apply EH to subset of markup language
GJR: EH needs to handle the contextualizing so it has to be aware of the ontology markup language using, so that it can convey info in proper context
NS: want to draw something HTML can't draw
MC: HTML doesn't have support for automatically numbered lists; can still use same accessible hierarchy as HTML
MC: for MusicXML can't be represented textually
NS: there is already a CSS profile for MathML
NS: the type of XML-derived languages we should be investigating are those that are represented by non-textual glyphs or are expressed in a non-linear way: MathML, SVG, MusicXML and CellML are most likely and useful candidate
GJR: i need to separate XML-wrappers from truly specialized content which poses rendering or input problems
ACTION GJR: separate XML-wrappers from truly specialized content which poses rendering or input problems
NS: say something meaningful about data - description of graphical relationships which could be described
MC: too tied up with XML here; going to parse, but isn't concern -- can we parsing output can't parse before?
MC: RixML interesting - does something HTML can't do, but output mechanism is still text
MC: MathML is a different output form
NS: agree that XML is a bit of red herring - an Open Office or Word document which a user needs rendered may need same expert handler
NS: all agreed on SVG and MathML and MusicXML as examples of need for expert handlers; question: are there more? what are common properties?
MC: physics markup language
NS: SVG, MathML, MusicXML, CellML, SBML
NS: concentrate on client-side rather than server-side
MC: currently, we can handle HTML, but not MathML i think the basic questions we need to ask and answer are:
MC: it is the on the client-side that we have to ask questions
NS: last week on the AIA call, there was some discussion with ZoomText about handling math and expert handlers; in my mind this is how i think it should go: suggested that EH registers with app, AT queries anyone handles node? if none handles, turns to expert handler -- if attempting to navigate, provides navigation; if attempting to get braille, get braille
PB: didn't come to conclusion - query interface and query service based on role or other info about object, query for functionality then call methods to get what needs
PB: role would make sense - attributes indicate additional information
PB: would want to break XML into different pieces - term, denominator, numerator - sub-components?
NS: want single universal handler
MC: problem with this is that EH have to know about potential forms of output; has to know how to handle speech, braille and magnification; any new output mechanism has to be added to all EHs;
NS: don't quite agree - may be no braille form for SVG;
MC: present in tactile or textual form
MC: if not specific output mechanism, use text; some things though can't be expressed in text
MC: want to avoid accessible hierarchies; need to provide method of navigating through elements -- could be done by AT, but EH would have to handle that in special way
NS: 2 models: 1) turn focus over to EH and let EH do it - one AT vendor said works for me, other said out of the question; not sure what is right solution; navigation commands -- generic commands that make sense to all EH then sub-class to music, math or whatever
PB: navigational means to move to parent and another method to put focus where wanted
MC: be problem for some ATs
PB: wouldn't be loosing control
NS: can pass event on after filtering it and asking want it or no?
PB: wouldn't turn focus over, but would enable setting of focus
MC: efficiency concern
PB: AT still calling SetFocus
MC: other suggestion is turn over tab navigation
MC: still wonder whether have to divide info into chunks, define accessible hierarchical ways to express this info; such a new area, don't know of ways can be done
discussion postponed until 24 November 2008
NS: schedule for next few weeks: